When I made the new domains test, one thing I wanted to see was whether people who are too into typology answer questions more unrealistically or not. My method was to create something called a “magic level,” which corresponds to particularly unusual or unrealistic behavior or thinking (e.g. Ni stare), and then include a question asking the test taker about how familiar they are with the cognitive functions. In the chart above, we see the averages corresponding to the different levels of familiarity (“I know nothing about [the cognitive functions]” at level 1 to “I’m an expert at [the cognitive functions]!” at level 4).

And now, performing both an ANOVA and Hukey HSD test (at a 99% cf. level) on the data, we’ll find that these groups are mostly statistically significant! Professed experts score higher on magic level than people who know nothing about them. The two groups that were not statistically significant were between level 1 (know nothing) and level 2 (know some things).

Edit 2023/12/1:
More data has come in, and the updated values seem to exacerbate these trends even further.
No specified level. Average 19.5088. Standard deviation 10.3783. 7k+ rows.
Level 1. Average 18.4305. Standard deviation 10.4217. 33k+ rows.
Level 2. Average 20.2346. Standard deviation 10.0958. 79k+ rows.
Level 3. Average 23.7587. Standard deviation 10.3088. 17k+ rows.
Level 4. Average 28.0390. Standard deviation 11.4511. 2k+ rows.